199 | P a g e
FEATURE EXTRACTION OF POWER SIGNAL
1
Aswini Biri,
2Anitha N,
3Upendra Naidu G,
4M.Vijay,
5Jaya Prakash A
1, 2, 3, 4
Dept of ECE, TPIST, Bobbili, AP, (India)
5
Assistant professor, Dept of ECE, TPIST, Bobbili, Komatipalli, AP,(India)
ABSTRACT
Our work provides an effective feature extraction of a power signal disturbances for analyzing both steady
state and short duration non-stationary signals. Power signal is a non-stationary signal in which statiscal
characteristics change with time. If u look at such a signal for a few moments and then wait for a period and
look at again it would not essentially be the same i.e. its amplitude distribution and standard deviation would
be different. In this project we investigate the features of different power signal disturbances like energy,
entropy, standard deviation etc. By using these features we can easily analyze the power signal disturbances.
Key words: Non-stationary power signal, feature extraction, 3-D feature plots, analyzing power signal
disturbances
I. INTRODUCTION
1.1 Non-Stationary Signals
A non-stationary signal is the one in which statistical characteristics change with time. If you look at such a signal for a few moments and then wait for a period and look it again, it would not essentially the same, i.e. its amplitude distribution and standard deviation would be different. Non – stationary signals can be divided into two types:
1.Mommentarily transient 2.Persistent
200 | P a g e
Fig 1.1 Non-stationary signal
One possible course of an action in such a case is to use a non-uniform sampling technique. The proper application of non-uniform sampling suppresses the frequency aliasing and allows the use of a sampling density below the Nyquist rate. It should be stated that non-uniformly taken signal require the focusing of more attention on the signal processing algorithm.
In communication systems accurate and quick measurement of signal frequency and amplitude in the presence of distortions and noise is a challenging problem encountered by communication engineers at present .Frequency and amplitude are important parameters for any communication signal quality monitoring, operation management and control using microprocessor. In the past various techniques were proposed that have been used in the estimation of signal frequency and amplitude perform well when the signal is sinusoidal in nature. But when random noise and harmonic distortions are present accuracy and convergence estimation are Discrete Fourier Transformation, Mean square Error, Kalman, adaptive notch filters etc.
II. GENERATION OF SYNTHETIC SIGNALS
2.1 Power Quality
Recently Power Quality (PQ) and related power supply issues have become quite a serious problem for both the end users as well as the utilities. According to established reports, the industry is losing huge amount of resources due to power outages and PQ problems. PQ may be defined as the continuous availability of electric power that confirms to accepted standards of phase, frequency, and voltage. Power System Disturbance is defined as a variation in these standards. The PQ issues and related phenomena can be attributed to the use of solid-state switching devices, unbalanced and non-linear loads, industrial grade rectifiers and inverters, computer and data processing equipments etc., which are being increasingly used in both the industry and home appliances.
These devices introduce distortions in the phase, frequency and amplitude of the power system signal thereby deteriorating PQ. Subsequent effects could range from overheating, motor failures, inoperative protective equipment to power inrush, quasistatic harmonic distortions and pulse type current disturbances. Therefore it is essential to monitor these disturbances. There are different types of disturbances which occur in power systems due to varied reasons. These disturbances can be broadly classified into two types.
Steady state disturbances
201 | P a g e
III. FEATURE EXTRACTION
In this work we have used two types of features i.e. time-domain and frequency-domain features. Out of the seven features used, spectral entropy is a frequency-domain feature and the rest, which includes energy, mean, standard deviation, variance, autocorrelation and maximum deviation, are time-domain features. From these features 5 statistics
f1… f5 are calculated which are fed to the network for classification task. All the time domain feature statistics
calculated are normalized between the ranges zero to one. This is done by dividing the individual features obtained for all the training patterns or (disturbance signals) with the maximum value of the respective feature. This ensures better stability of the network weights during training. There after the entropy value is normalized by dividing it by the factor log (N) where N is the total number of frequency components in the range [f1,…,fN]. It can be observed that the
entropy value for low frequency disturbances like voltage sag, voltage swell, momentary interruption and the pure undistorted sinusoidal is minimum, which is zero. Harmonics including sag with harmonics and swell with harmonics have a comparatively higher value of entropy while disturbances like voltage flicker and flicker involving harmonics show maximum entropy value in the class of steady state PQ disturbances.
IV. SIMULATION RESULTS
Energy plot:
0 50 100 150 200 250 300 350 400 450 500 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure 4.1 energy plot
0 50 100 150 200 250 300 350 400
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 TIME SAMPLES EN TR O PY
202 | P a g e
0 50 100 150 200 250 300 350 400
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
TIME SAMPLES
S
TA
N
D
A
R
D
D
E
V
IA
TI
O
N
Figure 4.3 standard deviation
Figure 4.4 plotting the entropy plots for different disturbances
203 | P a g e
Figure 4.6 plotting the standard deviation of the different disturbance signals
5. CONCLUSION
In this paper, an attempt has been made for pattern recognition and classification of power quality disturbances. In this we analyzed the time domain features like energy, standard deviation, mean, variance and frequency domain features like entropy of different power signal disturbances like voltage sag, swell, harmonics, momentary interruption etc.From that we can observe the different power quality disturbance signals within very short duration. And in future finally time – time transform has been utilized to produce representative feature vector plot that can capture the unique and silent characteristics of each disturbance. And further these values are applied to neural networks.
REFERENCES
[1] R.G. Stock well, L. Mansinha, and R.P. Lowe, Localization of the complex Spectrum:The S-Transform”, IEEE Transaction on Signal Processing, vol.44, No.4, 1996, pp.998-1001.
[2] C. R. Pinnegar and L. Mansinha “A method of time-time analysis: The TT-transform”, Elsevier Science on Digital Signal Processing, 13 (2003)588-603.
[3] Malay, and W.L.Hwang,” Singularity detection and processing with wavelets”.IEEE Trans. Information Theory, 38:617-643.1992
[4] D.Gabor, Theory of communications, J.Inst .Elec.Eng.,93(1946),pp429-457
[5] C. R. Pinnegar and L. Mansinha “A method of time-time analysis: The TT-transform”, Elsevier Science on Digital Signal Processing, 13 (2003) 588-603.
[6] Chu, P.C., the S-transform for obtaining localized spectral Mar. Technol.Soc.J., 29, 28-38.
[7] Mallet, S., 1998, a wavelet tour of signal processing: Academic Press. Mansinha, L., Stock well, R.G., and Lowe, R.P., 1997, Pattern analysis with two-dimensional spectral localization: Applications of two-dimensional S-transform: Physica A, 239, 286-295.
204 | P a g e
[9] Ronald R.Coifman and Malden Victor Wickerhauser “IEEE Transactions on Information Theory”, Vol 38, No 2, March 1992
[10] J.F Bercher and Vignat “Estimating the Entropy of a Signal with Application”Equipe Communications Numerous, ESIEE 93 162 Noisy-le-Grand, FRANCE and Laboratories Systems, UMLV.
[11] Birendra Biswal, A.Jaya Prakash,”A new approach to Time-Time Transform and Recognition of Non-stationary signals using FWNN
[12] A Jaya Prakash, S Daya Sagar Choudary, M S Vamsi Krishna, A L G N Aditya,”3-D Feature plots and pattern Recognition of various Non-Stationary Signals Using FWNN”.
Books
